When I was a graduate student at Cornell, I needed to go to the library with a reader (yes, an actual live human, not a computer:-)), find the relevant publications, and then arrange to have what I judged to be the most relevant articles read to me. Occasionally, I would scan the printed articles to OCR them and have my computer read it aloud to me. Mostly this was a failure since the articles I wanted to read were in the fields of math or computer science, and OCR dealt poorly, if at all, with technical material. More often than not, I would write to the authors of the publications in the hope of getting access to online versions so that I could save on reader time.
Turning the clock forward to today, equivalent access is a Google Scholar search away. What's more, thanks to publishers like the ACM who making journal publications available online, one can access the complete publication directly from within the search hits. Since the advent of Google Scholar, I have not had to send out email requests to authors asking for access to the electronic versions. And where authors have made these available from their websites, Google Scholar links to those as part of the result set.
So I wish I had had this as a graduate student. Though I have to say working at Google does feel like being in graduate school (but with better food and more money)—so maybe my wish has been granted.